[Axiom-developer] mnt binary storage

[C Y]

Tim, as long as I'm stepping on land mines here's one more to round out
the day:

As I mentioned before, I'm in the process of figuring out how to make
asdf handle pamphlet files.  Of course, this basically means taking
over as much of the build logic as possible, and one of the questions
that arises is where to stick .o or .fasl files (the binary results of
a lisp compile-file).

I know Axiom has a directory setup for intermediate files, and I'm
going to try to make sense of that and teach asdf to respect it, but I
wanted to ask you specifically abou the whole mnt/linux/... storage of
the final results.

There is an asdf extension called asdf-binary-locations 
http://common-lisp.net/project/cl-containers/asdf-binary-locations/
I'm thinking this might come in handy for the Axiom build, since we
wind up stuffing our binaries in different directories from the source
files.  It is configurable according to the documentation, but also has
defaults.  According to this post:
http://groups.google.com/group/comp.lang.lisp/browse_frm/thread/72296ec04e5cb356/8d98a545b85c03f6

this scheme is used to keep things straight:
root/compiler-name-version-os-arch/path/to/file/foo.fasl 

This seems to be a variation on what we have with the mnt setup.  My
question - is the mnt structure currently in place required/hard
coded/have overwhelming logic behind it?  I sort of like the
compiler-name-version-os-arch directory namimg scheme, as it should
make it trivially simple to compile Axiom for multiple lisps on one
machine (even multiple versions of the same lisp!) and let ASDF keep
everything straight.  Is that binary structure something I shouldn't be
monkeying with?  

Cheers,
CY


      
____________________________________________________________________________________
Shape Yahoo! in your own image.  Join our Network Research Panel today!   
http://surveylink.yahoo.com/gmrs/yahoo_panel_invite.asp?a=7